Fix validate_trace_empty test check
authorJulien Desfossez <jdesfossez@efficios.com>
Thu, 21 Dec 2017 20:28:56 +0000 (15:28 -0500)
committerJérémie Galarneau <jeremie.galarneau@efficios.com>
Wed, 25 Apr 2018 15:21:11 +0000 (11:21 -0400)
Since the output of babeltrace was directly piped into wc, the return
code was never an error even if the trace was invalid. We now split the
commands in two parts: process the trace with babeltrace and check the
error code, and then count the number of lines.

Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
tests/utils/utils.sh

index e8dfcda3cedf157c334d594bbc187d42020c4864..d8c8a1fed0e5dbd69c3b2d3a84e4779ea7a56e67 100644 (file)
@@ -1498,7 +1498,14 @@ function validate_trace_empty()
            skip 0 "Babeltrace binary not found. Skipping trace validation"
        fi
 
-       traced=$($BABELTRACE_BIN $trace_path 2>/dev/null | wc -l)
+       events=$($BABELTRACE_BIN $trace_path 2>/dev/null)
+       ret=$?
+       if [ $ret -ne 0 ]; then
+               fail "Failed to parse trace"
+               return $ret
+       fi
+
+       traced=$(echo -n "$events" | wc -l)
        if [ "$traced" -eq 0 ]; then
                pass "Validate empty trace"
        else
This page took 0.02565 seconds and 4 git commands to generate.